Multivariate forecasting in rapidminer

Barclaeys

Hello,

I am fairly new to Rapidminer and data science in general.

I am working on a small project in context of a postgraduate and I would like to build a time series model on monthly consumption data (dataset 1) but I would like to add another external source (dataset 2), also monthly aggregated, which influences the consumption.

I have seen this is called multivariate forecasting but how could I build such a model in Rapidminer?

Thanks beforehand,
Bart

lionelderkrikor

Hello @Barclays,

I think that one solution is : 
 -  FIRST to join your 2 datasets.
 - THEN to use the Windowing operator to transform your time series problem into  a "classic" Machine Learning problem.

To see and understand how to proceed, I encourage you to see the pedagogic video of @tftemme about the Windowing operator
in the RapidMiner Academy :

https://academy.rapidminer.com/learn/video/using-windowing-on-time-series-data?utm_source=studio

Hope this helps,

Regards,

Lionel 

Barclaeys

@lionelderkrikor thanks for replying Lionel. I'll take a look and let you know. Best, Bart

Pavithra_Rao

Hi @Barclays

Here are 2 Time Series webinars for your reference.

https://rapidminer.com/resource/time-series-foundations/
https://rapidminer.com/resource/advanced-time-series-forecasting/

Cheers,
Pavithra

Barclaeys

Hi @Pavithra_Rao, thank you so much. so far the webinars have been extremely useful and I have already been able to explore the foundations of time series. I'm now trying to extend to multivariate forecasting (with only one additional variable) but my dataset <100 rows hence I cannot use the automodel. Is there a way I can build a normal process like for basic forecasting but taking into account the multi-variate? Many thanks, Bart